Conductive Anodic Filamentation (CAF) is a failure mechanism which is found in Printed Circuit Boards (PCB). It is the creation of a micro conductive or semi-conductive path between two conductive poles. As shown in the photos below
These are thankfully rare but difficult to find and analyse – very much a case of finding a needle in a haystack.
What are the main conditions that could give rise to them.
In over 35years I have only ever seen four cases of inclusions of metal shards in the base laminate coming in from the suppliers, and one of carbon fibers these were in comparison easy to spot. CAF if it does happen is mainly down to a weakening of the material at the drilling stage which is then susceptible to chemical ingress during subsequent wet chemical processes.
The number of holes as a result of high interconnections in a design is directly proportion to the risk of CAF. It has been said that some boards are more hole than material. Which as any mechanical engineer will tell you alters the physical properties of the material.
By simple analysis a CAF can only occur between two conductive paths in a horizontal plane which is at 90 degrees to the direction a hole is drilled. A pcb is made up of horizontal layered parts forming a matrix with variable tensile and compressive strengths within the structure, but with predictable “composite” strengths.
Drilling of PCB’s is a tried and tested formula with given drill speeds for a given hole size. The only thing that changes is what you are drilling. Every design will present a different variation nevertheless a general approach can to be taken rather than setting parameters for every separate hole.
Figure 2 shows in part some of the more complicated builds that are being made today, each and every type of hole gives a different challenge to the drilling department, who in the real world have limited control options.
To understand CAF’s you have to look at the micro dynamics within a drilled hole.
As the drill enters the material it is faced with copper , a soft malleable metal, once it has cleared this it hits fiberglass and resin which will vary in hardness and malleability depending on the resin type and content of any given point. Then it hits copper again. This is repeated depending on how many inner layers there are.
The drill will have to cut or chip its way through between 2000 and 10000 holes or hits: this is the equivalent of 3 to 16 meters of pcb. The cutting edge is rotating at anything up to 200 thousand RPM so a 0.2mm drill will see 105mm of material per second and 2Km before it is recommended to change it.
There are a few different aids to drilling : the correct choice will have a great bearing on the quality of holes.
1)Entry board: Helps with setting the drill and preventing drill deflection.
2)Exit board : Helps prevent burrs and drill damage against hard drill beds.
3)Pressure foot settings
There are other potential factors such as chip loading, entry speeds withdrawal speeds, and are dependent on drill size and are normally preprogrammed machine settings.
CAF’s do not appear at drilling, but the micro fissures that can give rise to CAF’s do.
The glass filaments are chipped rather than cut so there are always fibers that are pulled out of the surface of the hole wall. It is in fact desirable to have some surface roughness for the subsequent processes. Benefits include adhesion and propagation.
Desmear is the process by which resin that has re-deposited over an interconnection of copper is removed.
There are two basic methods. Wet chemical or Plasma gas.
Both do effectively the same thing the oxidation of organic carbon based material (epoxy) without effecting the copper or fiberglass . At the Desmear stage only the surface of the hole wall and panel edge will be exposed to the treatment. Plasma is a cleaner process but considerable more expensive and is only done by batch method. Permanganate or wet chemical desmear has variable effects on different resin systems, and doesn’t work at all in some scenarios (such as high aspect ratios – air traps, polyimide designs).
Resin Smear over interconnect normally removed by desmear.
A multilayer construction can vary in the number of bonds a finished board can undergo.
By looking at the above example again there are at least three separate bonding cycles but at least 7 separate drilling stages.
The buried vias on layers 4-13 would be bonded and drilled under one set of parameters. The layers 4 and 13 would be all copper at the desmear stage. These would then be imaged, plated, etched and optically and or electrically tested for shorts.
The Stacked holes from layers 1-3 and 16 to 13 could have been drilled using several different methods. Depending on the size and clearance tolerances available.
Buried micro-vias tend to be drilled using laser techniques.
There is again an element of complication, as the layer count mounts up the thickness of the inner layer decreases. These have to go through the same chemical processes as a rigid board and can be like processing tissue paper. They are more prone to handling and machine processing damage. These range from minor scratches to major factures in the glass epoxy matrix, and thereby increase the possibility of a CAF occurrence. Normally these are seen as white spots in the laminate and become the prime sites for the formation of CAF’s. There is a limited degree of self repair during the bonding process as the pre-pregs flow to fill in damaged areas, but where this is incomplete or insufficient CAF may occur.
It is easy to see that even with the greatest care that can be applied to the processes the potential for a CAF to form at any stage may be dependent on processes that were several stages before and not at the final stages.
The last holes to be drilled are the through hole vias. These physically pass through material that has had multiple wet processes applied to it along with, in the above case, several bonding cycles with heat and pressure. If these pass through any weakened areas then this is where a greater propensity for a CAF to form occurs.
These are microscopic failures which even with the best Micro-sectioning cannot pick up. They are not to be confused with de-lamination and blistering, which are different failure modes.
Once the through hole has been drilled the panel will be subjected to yet another desmear process.
The picture on the left shows good micro-roughing of the resin system and the one on the right shows good smear free copper inner layer ideal for successful metallization of an interconnection after de smearing.
The above photo shows an example of drill ploughing where the drill will cut an uneven path as it passes down the hole. Giving the appearance of a furrow having been made.
A worn drill gives rise to rougher drilled holes as the cutting edge loses its chipping ability and a higher degree of lateral pressure. Blocked or damaged drills, which does happen even with broken drill detectors, also give rise to rough holes.
As with inner layer thickness the pre-pregs also become thinner with higher layer counts.
Prepregs (think glue layers used to stick the inner conductive layers together) can also become damaged through bad handling techniques which in-turn can give damaged fibre cloth, which increases the possibility in turn of CAF.
Different types of PCB material will have different properties. As a general rule higher TG (Glass Transition Temperature) materials are harder in nature and tend to be tougher to drill and desmear. Mixing laminate types within a build and introducing thermal cladding can be challenging but not insurmountable. The exception being PTFE type material s which are conversely as soft as putty: and have to be treated in a very different manner.
Then there is the whole area of oxide treatment of the copper layers prior to bonding.
This area is historically well known for de-lamination and blistering problems. Fortunately process and material technology has improved significantly over the last decade, and if carried out correctly these issues are thankfully very rare.
The process basically micro roughens the copper surface along grain boundary lines. This increases the surface area to be bonded by a factor of up 10 times, which leads to improved adhesion and hole strength integrity. Originally this copper was oxidized and had a Black oxide applied which was susceptible to acid ingress seen as “ Pink ring”. This Pink Ring used to be a serious scourge in the industry – (this was where an area of Black Oxide had delaminated from the side of the treated hole and subsequent acid processing interacted at this interface and oxidised the materials in the opening resulting in delamination of the hole plating, and would appear as a Pink Ring).
To overcome this issue a reduction process was introduced which reduced the oxide back to metallic copper but still retained the colour of the Black Oxide. Although this over came the Pink Ring issue it could lead to any copper salts present being reduced back to Metallic copper which in turn if left on the surface of an inner layer would have been a perfect site for CAFs to form.
Technology has moved on again and a newer generation of Brown oxides do not exhibit the same flaw. However as the working solution in the treatment bath ages, the buildup of copper compounds in the bath can precipitate onto the surface of inner layers which if not effectively rinsed could potentially lead again to CAFs. As always its a fine balancing act.
However, the Brown Oxide process is yet another chemical process where copper salts can be adsorbed onto a surface or trapped in micro fissures. This has been shown to be the case with any wet chemical solution or mechanical brushing process.
The separate stages of manufacture expose the surfaces to many different acids, alkalis ,oxidizing and reducing chemicals.
The presence of moisture will make salts conductive (again leading to CAF). However, reduction processes can reduce salts to base metal giving rise to conductivity (also leading to CAF). The golden rule is cleanliness is next to Godliness, and processes should not be pushed beyond their stated parameters – tolerances and SPC (Statistical Process Control) of process baths to optimum operational effectiveness . Even the mildest of cleaners will pick up copper which can be deposited on to a surface and if allowed to dry can deposit copper salts resulting in CAF.
These may not become a CAF until they are drilled into and a connection is made, but the opportunity is now present.
They may not ever become a problem but when they do they can be catastrophic.
There is no one cause for CAF’S, as they can be the result of any combination of different factors.
The underlining facts are that there has to be a physical weakness in the material caused by either drilling or handling that allows for the ingress of copper salts from wet chemistry processes for there to be a potential site that then connects between two conductive parts.
If all good housekeeping rules are followed and any defects are seen and rejected at the various inspection stages, then the incidence of CAF is reduced to a bare minimum.
Pile up the layers, Pile up the hole count pile up the pressure to get the board made even faster, over engineer the build and buy at the lowest price then don’t be too surprised when there is more than one Needle in that Haystack.